Determining the factors which control deep convection over tropical oceans has historically been one of the most difficult problems of meteorology. In this talk I outline progress on this problem which has been made in the past decade as the result of theoretical developments plus field programs such as TEXMEX, TOGA COARE, EPIC2001, KWAJEX, etc.

On small space and time scales, mechanical forcing (gust fronts, boundary layer turbulence, etc.) of convective parcels is important.On larger scales thermodynamic forcing dominates, with the only dispute being the scale at which the transition from mechanical to thermodynamic forcing occurs. Theoretical arguments suggest that the transition occurs at scales of a few hundred kilometers and about one day. I deal only with thermodynamic forcing here.

Thermodynamic factors that appear to promote deep convection and in particular strong convective rainfall are

1. The existence of a moist free troposphere -- precipitation in cumulus ensemble models is exquisitely sensitive to this.

2. Strong surface heat fluxes produced by a combination of high SSTs and strong surface winds. The strongest fluxes generally occur upstream of the convection.

3. The absence of a capping inversion near the 800 mb level.

Incorporation of these sensitivities into cumulus parameterizations in large-scale models is likely to be important to obtaining a satisfactory representation of convection and precipitation over tropical oceans.